Empirical prediction of resistance of fishing vessels

Abstract

The possibility to use Artificial Neural Network for estimating ship resistance and propulsion coefficients are investigated. Different ANNs are tested by varying input parameters, network size and complexity and division of data material into training and testing sets. ANN prediction methods are trained for Resistance (Cr), total propulsion efficiency (nD), open water efficiency (n0), hull efficiency (nH), wake fraction (w), thrust deduction (t) and relative rotative efficiency (nR). The data material for the thesis are model test results from MARINTEK and consist of 193 fishing vessels and loading conditions.